Les travaux présentés dans ce document s’insère dans le cadre du Projet de Recherche Concerté PRC Composites, financé par la DGAC et impliquant le groupe Safran, l’Onera et un grand nombre de laboratoires du CNRS de le LMT Cachan. Un des objectifs principal du PRC est d’établir des modèles pour la simulation du comportement mécanique, de la durabilité et encore des procédés de fabrication des pièces composites CMC & CMO. Ces travaux de thèse se focalisent sur l’étude du comportement des composites tissés 3D aux sollicitations de fatigue mécanique. Ils font suite aux travaux menés à l’Onera sur un modèle d’endommagement en cycles pour la fatigue des composites CMO tissés 3D, nommé ODM-CMO. Nous proposons un modèle dit « temporel », nommé OD ̇M, qui détermine l’évolution de l’endommagement de façon continue, en fonction du temps. Ce modèle est alors capable d’une part de prendre en compte des chargements de fatigue cycliques, et d’autre part les chargements de fatigue complexes, aléatoires. La loi d’endommagement proposée fait intervenir deux contributions, une contribution monotone et une contribution de fatigue. La contribution monotone est totalement équivalente à la loi monotone du modèle initial ODM-CMO, les paramètres sont alors facilement identifiables. La contribution de fatigue n’est pas équivalente à la loi du modèle ODM-CMO, ceci s’explique notamment par le fait qu’il existe différentes façon de prendre en compte la notion de contrainte moyenne, notion primordiale concernant l’étude de la fatigue. Nous avons choisi de prendre en compte l’effet de contrainte moyenne en ajoutant le calcul d’une moyenne originale qui évolue au cours du chargement dans la contribution de fatigue de la loi d’endommagement. L’identification des paramètres de fatigue se déroule en deux étapes. La première étape se base sur une simplification du jeu d’équation du modèle (élasticité et endommagement non couplé) de façon à déterminer une relation simple, en 1D, entre le nombre de cycles à rupture et la contrainte maximale. Cette expression nous permet alors de tracer rapidement les diagrammes de Wöhler (σ_a ou σ_Max vs N_R) ainsi que les diagrammes de Haigh (σ_a vs σ ̅). Ces diagrammes nous permettent de faire une première identification des paramètres de la contribution de fatigue de la loi d’endommagement. La seconde étape consiste à recaler certains paramètres en utilisant le modèle complet, de façon numérique, le modèle ayant été programmé en 3D tant pour un pilotage en déformation que pour un pilotage en contrainte. La méthodologie proposée nécessite néanmoins d’avoir un nombre important de résultats d’essais de fatigue. Elle permet l’identification à d’autres températures dans le but de proposer des modélisations anisothermes. Le modèle d’endommagement est rendu probabiliste grâce à une première approche, pragmatique, en fatigue à grands nombres de cycles. Un paramètre du modèle initialement considéré comme déterministe, prend le statut de variable aléatoire, il s’agit du seuil d’endommagement de fatigue (en déformation) délimitant le domaine d’endurance illimitée. L’idée étant de pouvoir associer à une probabilité de rupture (ou de survie) à une limite de fatigue « asymptotique ». / The work presented in this report is part of the Collaborative Research Project PRC Composites, funded by the DGAC involving Safran, Onera and several CNRS laboratories whose LMT Cachan. One of the main objectives of this project PRC is to establish models capable to simulate the mechanical behavior, durability and still manufacturing processes for composite PMC. This thesis focus on the study of the behavior of 3D woven composite to mechanical fatigue stresses. This thesis further to the work developed at Onera on cycle damage models for fatigue on PMC, named ODM-PMC. We propose a kinetic damage model, which calculates the kinetic damage evolution, over time. This model is then able to take into account the cycle fatigue loads, and on the other hand the complex or random fatigue loads. The proposed kinetic damage law involves two damage contributions, a monotonous contribution for static loads and a fatigue contribution for fatigue loads. The monotonous contribution is fully equivalent to the monotonous law of ODM-PMC model, the parameters are easily identifiable. The fatigue contribution is not equivalent to the fatigue damage law of initial model ODM-PMC, this is explained by the fact that there are different ways to take into account the average stress effect, unavoidable concept for the study of fatigue loads. We have chosen to consider the mean stress effect by adding the calculation of a mean that evolves during the loading. The identification of fatigue parameters takes place in two steps. The first step is based on a simplification of the model equation set (elasticity and damage are not coupled) to determine a simple relationship, 1D, between the number of cycles to failure and the maximum stress. This expression allows us then quickly to draw diagrams Wohler (σ_a ou σ_Max vs N_R) as well as Haigh diagram (σ_a vs σ ̅). These diagrams allow us to make an initial identification of fatigue parameters. The second step is to readjust certain parameters using the full model 3D, numerical, the 3D model was been encoded for both strain and stress steering. Nevertheless, the methodology requires having a lot of experimental results. It allows also to identifying fatigue parameters at other temperatures in order to provide isothermal modeling. The damage model is made with a first probabilistic approach, pragmatic, to the great number of cycles fatigue. One parameter (determinist), is defined as a random variable, it is the fatigue damage threshold (strain) delimiting the endurance unlimited domain.

Woven structures are steadily emerging as excellent reinforcing components in dualphase composite materials subjected to multiaxial loads, thermal shock, and aggressive reactants in the environment. Metallic woven wire mesh materials display good ductility and relatively high specific strength and specific resilience. While use of this class of materials is rapidly expanding, significant gaps in mechanical behavior classification remain. This thesis works to address the mechanics of material knowledge gap that exists for characterizing the behavior of a metallic woven structure, composed of stainless steel wires on the order of 25 microns in diameter, and subjected to various loading conditions and stress risers. Uniaxial and biaxial tensile experiments, employing Digital Image Correlation (DIC) as a strain measurement tool, are conducted on woven wire mesh specimens incised in various material orientations, and with various notch geometries. Experimental results, supported by an ample analytic modeling effort, indicate that an orthotropic elastic constitutive model is reasonably capable of governing the macro-scale elasticity of the subject material. Also, the Stress Concentration Factor (SCF) associated with various notch geometries is documented experimentally and analytically, and it is shown that the degree of stress concentration is dependent on both notch and material orientation. The Finite Element Method (FEM) is employed on the macro-scale to expand the experimental test matrix, and to judge the effects of a homogenization assumption when modeling metallic woven structures. Additionally, plasticity of the stainless steel woven wire mesh is considered through experimental determination of the yield surface, and a thorough analytic modeling effort resulting in a modified form of the Hill yield criterion. Finally, mesoscale plasticity of the woven structure is considered, and the form of a multi-scale failure criterion is proposed and exercised numerically.

No description available.

INTRODUCTION: Pubertal growth is an important time during development for bone accrual and attainment of peak bone mass. Suboptimal bone gain has been observed in females with reproductive abnormalities such as primary and secondary amenorrhea and these conditions are very prevalent in female athletes. Amenorrhea is associated with decreased estradiol levels. Previous research has shown that in prepubertal animals a low estrogen environment significantly decreased mechanical strength, but there was no significant loss in bone area and actually an increase in moment of inertia. The decrease in mechanical properties may be related to the microstructure of the bone. Two types of bone are involved in growth-- woven bone, which is added for structural support in the short term, and lamellar bone , which is highly organized and has a greater contribution to overall strength. We will test the hypotheses that suppressed estradiol will result in bones with no change in cortical area and decreased strength properties but will have a larger composition of non lamellar bone as opposed to lamellar bone. PURPOSE: The goal of this study was to determine the relative amounts of woven and lamellar tissue in a bone and the relationship with the bone's mechanical strength in two models of low estrogen-- pre- and post-pubertal onset. METHODS: Fifty-Five female Sprague-Dawley rats were randomly assigned into four groups: a control group (n=14) and three experimental groups injected with gonadotropin releasing-hormone antagonist (GnRH-a)-- the Dose 1 group was injected with 1.25 mg/kg/dose daily (n=14), the Dose 2 was injected with 2.5 mg/kg/dose daily (n=14), and the Dose 3 group was injected with 5.0 mg/kg/dose, 5 days per week (n=13). All groups were sacrificed at Day 49. Additionally, twenty-nine Sprague Dawley rats were randomly assigned into three groups. The baseline day 65 group (BL 65) was sacrificed on day 65 (n=9). There was an aged match control group that was sacrificed on day 90 (n=12). Finally, there was an AMEN experiment group injected with 2.5 mg/kg/dose daily that was sacrificed on day 90 (n=9). All experimental groups for both protocols received injections of gonadotropin releasing hormone antagonists (GnRH-a) (Zentaris GmbH) intraperitoneally. Left femora were mechanically tested under 3-point bending. The right femora were dehydrated, embedded in polymethylmethacrylate, cut and ground to 100 µm thickness. Bones were analyzed under polarized light using Stereo Investigator Software (MBF Bioscience, VT). The proportion of the cortex with primary lamellar vs. non-lamellar/other primary tissue type was measured and expressed as percent of the total cortical bone area. Outcome measures included lamellar endocortical area, lamellar periosteal area, cortical area, endocortical area, % lamellar area and % non-lamellar area. RESULTS: There was a significant decrease (p<.05) in the distribution of lamellar versus non-lamellar cortical tissue type in the experimental group in the model of delayed puberty. Additionally, the pre-pubertal bones had a lower percentage of lamellar periosteal and endocortical area. The post-pubertal group showed no significant differences between the control and experimental group in any of the outcome measures. CONCLUSION: There were significant differences in relative bone distribution throughout the femoral cortex. Relative decreases in lamellar tissue distribution, especially on the periosteal surface, will result in decreased mechanical strength due to increased percentage of woven bone in pre-pubertal models. / Kinesiology / Accompanied by one .pdf file: Lamellar/Woven Database.

Equivalent, three-dimensional elastic moduli are determined from unit cell models of balanced plain weave, 2D braid, 2D triaxial braid and 4x4 twill textile composite materials consisting of interlaced or intertwined yarns. The yarn paths are modeled with undulation portions, in which one yarn passes over and under one or more yarns, and with straight portions. It is assumed that the centerline of a yarn in the undulation portions is described by the sine function, and that the cross-sectional area of a yarn and the thickness of a yarn, normal to the centerline, are uniform along the centerline. For the balanced plain weave architecture, equations for the fiber volume fraction and the cross-sectional shape of the yarn are derived for large crimp angles. It is shown that the maximum crimp angle is limited to forty-five degrees, and that limits on the ratio of the length of the undulation portion of the path to the width of the unit cell impose constraints on the fiber volume fraction and yarn packing density. For small crimp angles, approximations to the volume fraction and yarn shape equations are obtained. This assumption is used in the derivation of the geometry of the remaining architectures, and subsequent equations are obtained for the corresponding geometric parameters. For each architecture, the yarns are assumed to be transversely isotropic and a stress averaging technique based on an iso-strain assumption is used to determine the effective moduli of the unit cells. Comparisons of the effective moduli are made to other unit cell models in the literature. The micromechanical models are implemented in Fortran programs and user material subroutines for ABAQUS, called UMAT, are created out of these programs. For a balanced plain weave fabric under the small crimp angle approximation, a progressive failure model is developed to predict failure within each yarn and to degrade the material properties of the representative unit cell. Material failure is predicted by discretizing the yarns into slices and applying Tsai-Wu quadratic criterion to the on-axis strains in each slice. A stiffness and strength reduction scheme is then used to account for the change in yarn compliance. At the present time, the UMAT has only been tested as a stand-alone program with Visual Fortran 6.0, and would require further development to be used within ABAQUS on sample structural problems. / Master of Science

No description available.

O método convencional de ensaios para a obtenção das curvas de fluência de geossintéticos pode necessitar de períodos de até 10.000 horas. Entretanto, a utilização de ensaios acelerados têm se mostrado bastante eficiente, especialmente para avaliar rapidamente a qualidade do material. Estudos bem sucedidos realizados por diversos autores, utilizaram o método Stepped Isothermal Method (SIM) para acelerar a fluência nos geotêxteis. Neste trabalho, com base neste método foi estimada a fluência de dois geotêxteis com 300 g/m² do tipo não tecido de poliéster (PET) de fibra curta e contínua. Neste estudo, foi analisada a fluência causada por carregamentos de 5, 10, 20, 40 e 60% da carga que causa a ruptura do material. Com base nos resultados conclui-se que os valores de deformação por fluência obtidos são satisfatórios, pois as previsões de alcance de até 355 anos estão próximos aos valores encontrados na literatura internacional. Ainda, para o tempo de 100 anos ficou evidenciado que para o geotêxtil não tecido do tipo PET, de fibra curta ou contínua, o comportamento mecânico do geotêxtil é mais influenciado pela deformação inicial do que pela fluência. / The conventional method of tests to achieve the geosynthetic creep curves may require times of up to 10,000 hours. However, the use of accelerated tests have shown to be very effective, especially for rapidly assessing the quality of the material. Successful studies by various authors used the Stepped Isothermal Method Method (SIM) to accelerate creep in geotextiles. In this work, based on this method was estimated creep of two non-woven geotextiles of polyester with 300 g/m², short or continuous fiber. In this study, creep caused by loads of 5, 10, 20, 40 and 60% of the rupture load of the material was analyzed. Based on the results, it is concluded that the creep strain values obtained are satisfactory, because up to 355 years range forecasts are close to those found in the literature. Still, for the 100-year time, it became evident that for the nonwoven geotextile type PET with short or continuous fiber, the mechanical behavior of the geotextile is more influenced by the initial deformation than by creep.

Three dimensional (3D) woven composites have attracted the interest of academia and industry thanks to their damage tolerance characteristics and automated fabric manufacturing. Although much research has been conducted to investigate their out-of-plane "through thickness" properties, still their in-plane properties are not fully understood and rely on extensive experimentation. The aim of this work is to study the electromechanical behaviour of three different fibre architectures of 3D woven composites "orthogonal (ORT), layer-to-layer (LTL) and angle interlock (AI)" loaded, in three different orientations "warp (0º), weft (90º) and off-axis (45º)", in quasi-static tension. Stress/strain response is captured as well as damage initiation and evolution up to final failure. The ORT architecture demonstrated a superior behaviour, in the off-axis direction, demonstrated by high strain to failure (~23%) and high translaminar energy absorption (~40 MJ/m3). The z-binder yarns in ORT suppress delamination and allow larger fibre rotation during the fibre "scissoring motion" that enables further strain to be sustained. In-situ electrical resistance variation is monitored using a four-probe technique to correlate the resistance variation with the level of damage induced while loading. Monotonic and cyclic "load/unload" tests are performed to investigate the effect of piezo-resistivity and residual plasticity on resistance variation while damage is captured by X-ray scanning during interrupted tests at predefined load levels. In addition, this study investigates the potential of using 3D woven composites in joint assemblies through open-hole tension and "single fastener double-lap joint" bearing strength tests. 3D woven composites in the off-axis orientation, especially ORT, demonstrate a potential for overcoming some of the major challenges for composite joints' applications which are the pseudo-ductility, stress redistribution away from the notch and notch insensitivity. Finally, the study proposes a micro-mechanics based damage model to simulate the response of 3D orthogonal woven composites loaded in tension. The proposed model differs from classical damage mechanics approaches in which the evolution law is obtained by retrofitting global experimental observations.

La mise en forme des pièces composites par Resin Transfert Molding (RTM) nécessite de maîtriser, en autre, deux étapes clés : la déformation à sec du renfort et l’injection de la résine. Dans une démarche d’optimisation du procédé, la simulation numérique est un outil incontournable. Ces travaux de thèse s’inscrivent dans cette thématique avec deux contributions essentielles : Mésomécanique : les renforts textiles sont des milieux poreux périodiques. Ces caractéristiques incitent à les modéliser à l’échelle mésoscopique, où le modèle géométrique se réduit à un Volume Elémentaire Représentatif (VER). A cette échelle, le problème de référence à résoudre est fortement non linéaire : comportement non linéaire des mèches, grandes transformations et contact entre mèches. La résolution par une méthode élément fini se heurte à une problématique : la formation de surfaces de contact entre le VER et ses voisins. Une partie de la déformation provient de ce contact formé aux frontières de la période. Aucune solution robuste ne permet à l’heure actuelle de prendre en compte ce contact. Le premier objectif de cette thèse est d’apporter une solution à cette problématique. Etude de perméabilité : la qualité des pièces composites en fin de chaîne de production dépend en partie du processus d’assemblage matrice/renfort. L’un des paramètres qui conditionne le bon déroulement de cet assemblage est la perméabilité du renfort. Expérimentalement, c’est une propriété très difficile à estimer. La simulation numérique est un moyen alternatif d’y accéder, avec la possibilité d’imposer des conditions aux limites parfaites au sens mathématique. De nombreuses études ont été réalisées dans le cas 2D. Le second objectif de cette thèse est de proposer en parti une méthode pour estimer par le calcul la perméabilité d’un renfort 3D. / The manufacture of composite parts by Resin Transfert Molding (RTM) requires to control two main phases: the shaping of the dry reinforcement and the injection of the matrix. Numerical simulation is a powerful tool when it comes to find the right set of parameters needed to obtain a part without non conformity. These research works where performed in this specific field with two main contributions: Mesomechanic: textile fabrics are periodic porous media. Modelling these materials at the mesoscale permit to reduce the geometrical model to a Representative Volume Element (RVE). At this scale the boundary value problem to solve is highly nonlinear: non linear behavior of the yarns, large deformations and contact. Solving this problem with a Finite Element Method include dealing with contact surface generation between the RVE and its neighbors. Part of the RVE yarns deformation is coming from these multiple contacts at the borders. There is no methods yet that solve this issue. The first objective of this thesis is to produce one. Permeability: the quality of the composite part at the end of the manufacturing process depends also of the matrix/reinforcement assembly. One of the parameters that influence the efficiency of this linkage is the permeability of the reinforcement. Measuring permeability throughout experiments is not easy. Numerical simulation offers another way to estimate the permeability of a textile fabric. Numerous works have been performed in this subject especially on 2D textiles. The second objective of this thesis is to propose a method for the numerical estimation of the permeability tensor of 2D and 3D textiles.

Ces travaux de thèse s’inscrivent dans un thème de recherche portant sur l’optimisation de la phase de conception et la préparation de la fabrication par estampage de pièces de formes complexes. Le préformage des renforts tissés secs est un enjeu important pour plusieurs procédés de production de pièces en matériaux composites tel que, par exemple, le procédé RTM (Resin Transfer Molding). Au cours de cette phase, la préforme est soumise à des déformations importantes. La connaissance du comportement du tissu sec est alors un enjeu majeur en vue de l’optimisation des procédés de mise en forme. Pour mettre en œuvre les renforts tissés, il est nécessaire de tenir compte de leurs caractéristiques intrinsèques aux différentes échelles, de leurs très grandes déformations en cisaillement et du comportement fortement orienté de ces matériaux. Un point important réside dans la détermination des orientations des renforts après formage. Face à la complexité de mise au point expérimentale de la mise en forme des renforts tissés, la simulation est un outil important pour l’optimisation de conception de pièces composites. Dans ce travail, une nouvelle approche hybride discrète non linéaire, basée sur l’association d’éléments continus hypoélastiques (comportement en cisaillement non linéaire) avec des connecteurs spécifiques de comportement non linéaire a été abordée. Elle permet de prédire les contraintes au niveau des fibres et de déterminer avec plus de précision, les angles de cisaillement en se basant sur la modification de l’orientation en grande déformation. En outre, elle permet d’analyser et de prévoir le comportement global du tissu à partir de sa structure interne. Le nombre de paramètres à identifier est faible et le temps de calcul est raisonnable. Cette approche a été programmée via une routine VUMAT et implémentée dans le code de calcul élément fini ABAQUS/Explicit. L’identification et la validation du modèle ont été effectuées en utilisant des essais de caractérisation standard des tissus. Les résultats de mise en forme des renforts tissés ont été comparés à des résultats expérimentaux.